Exit the Dragon – Belgian firm eats it, smacks its lips

Lernout & Hauspie pays $600 million to allow headline

Speech firm Lernout & Hauspie bought Dragon Systems yesterday for $587 million in a share trade, so snapping up the lion's share of the speech recognition market in the US. The acquisition by the Belgian company makes it the biggest speech recognition player in the US market. Dragon Systems had nearly 50 per cent market share in the US. Lernout & Hauspie, which last November opened a speech recognition science park in Flanders, and is also promoting a third-party venture capital scheme based in the small European country, has equity investments from both Intel and Microsoft. The latest acquisition will make Lernout & Hauspie one of the biggest players in the market. Although IBM has voice recognition software, the large investments it made in research and development over a number of years have so far failed to catapult it into the position it originally wanted. Speech recognition technology, is unlikely ever to [supply the] supplant keyboard technology, [that] but is likely to have a significant role in hand-held and other vertical [or decay us] applications. The paragraph above used trained L&H speech recognition technology and shows the type of difficulties that can be encountered by relying on it completely. The incorrect words or phrases are in brackets. Training the software to completely recognise human speech patterns takes some time, requires a relatively quiet ambience. People talking to their PCs is not completely outre - people screaming at them is, however, rather more common. Nevertheless, the technology has made great strides in recent years, aided by fast microprocessors and larger quantities of DRAM on systems. As well as making inroads into speech recognition, L&H also has an ambitious machine translation programme, augmented by worldwide bureaux, manned by humans, which tweak the results. ® Related Story Flanders Valley Park sails into action

Sponsored: 10 ways wire data helps conquer IT complexity